全球变化知识资源中心

Projections of ecosystem and biodiversity change for Africa under climate change diverge widely. More than other continents, Africa has disturbance-driven ecosystems that diversified under low Neogene CO₂ levels, in which flammable fire-dependent C₄ grasses suppress trees, and mega-herbivore action alters vegetation significantly. An important consequence is metastability of vegetation state, with rapid vegetation switches occurring, some driven by anthropogenic CO₂-stimulated release of trees from disturbance control. These have conflicting implications for biodiversity and carbon sequestration relevant for policymakers and land managers. Biodiversity and ecosystem change projections need to account for both disturbance control and direct climate control of vegetation structure and function.

African ecosystems and biodiversity are biologically and ecologically unique, attract substantial tourism revenue, and provide significant ecosystem services at local, regional and global levels. It is projected that anthropogenic climate change is likely to have adverse impacts on African ecosystems and their biodiversity¹, but projections of impacts based on a range of methodologies diverge widely. This is partly due to contrasting scenarios of future precipitation, but much more importantly due to critical differences between approaches to modelling biodiversity impacts and their assumptions. These differences relate to the extent to which different modelling approaches incorporate the effects of atmospheric CO₂ and disturbance (fire, mammal herbivory) on ecosystem structure and productivity^{2, 3}, and relative strengths in accounting for temperature- versus water-related controls on biodiversity. Projections of large declines in biodiversity under combined scenarios of climate and socioeconomic development^{4, 5} may not take these uncertainties into account. The most recent report by the IPCC on African vegetation change¹ reflects this well in stating with high confidence that “substantial uncertainties are inherent in these projections [future changes in terrestrial ecosystems] because vegetation across much of the continent is not deterministically driven by climate alone”.

These issues require urgent resolution, because there are significant and immediate implications for biodiversity risk assessments, and adaptation and mitigation responses relevant for policymakers and land managers. The evidence necessary to resolve these divergences is indicative but far from adequate, primarily because of a dire lack of empirical evidence and information on both climate and atmospheric CO₂ impacts on the structure and function of water-limited and disturbance-dependent ecosystems in tropical and subtropical climates^{6, 7}.

More than any other continent, Africa's ecosystems are water limited⁸ and disturbance driven (by wildfire and mega-herbivores)⁹, with a high representation of C₄ grass-dominated ecosystems^{10, 11}. Palaeoecological changes in climate and atmospheric CO₂ since the Miocene have strongly shaped the vegetation, disturbance regimes and biodiversity of these ecosystems¹². Because the continent straddles the tropics and subtropics in both hemispheres, an enormous area over which climatic conditions are conducive to vegetation flammability emerged during the late Miocene¹³. The palaeoecological factors that led to the dominance of C₄ grasslands under these conditions are complex and possibly regionally specific¹⁴, but after the prevalence of C₄ grassland systems gradually increased under the low atmospheric CO₂ of the early to middle Neogene¹⁵, the ecological ascendancy of flammable grassland systems was triggered by increasing climate seasonality¹⁶, and possibly entrained by reinforcing feedback between regional climate, vegetation and fire¹⁷. Africa's land surface today accounts for more than half the annual burnt area of the world^{18, 19}. Until a few centuries ago, Africa was also home to a near-intact megafauna, now becoming increasingly restricted to protected areas.

The late Miocene spread of grasslands provided novel habitats that supported high densities of grazing mammals, especially in Africa²⁰. This included the evolution of mega-herbivores with very substantial impacts on vegetation structure and function⁹. Changes in vegetation structure and disturbance regime in the late Neogene led to the spread of disturbance-dependent and disturbance-enhancing C₄-dominated grasslands and flammable shrublands, at the expense of closed forests, that today cover more than 70% of Africa's vegetated surface¹⁸. This is the largest global anomaly to the assumption that world vegetation patterns are primarily determined by climate²¹.

Disturbance-dependent African vegetation, including iconic savannas, has phylogenetic, functional and biodiversity characteristics that today are highly distinct from disturbance-sensitive biomes such as more ancient tropical forests^{11, 22}. Disturbance by fire and herbivores both limit tree cover²³ that to a first approximation differentiates plant forms that are tolerant of disturbance from those that are disturbance sensitive (although we acknowledge important differential aspects between these disturbance types that are less relevant for the purpose of this Perspective⁹). Importantly, these biomes are metastable, with interactions between disturbance, atmospheric CO₂ and climate drivers able to shift vegetation structure and composition rapidly²⁴, with strong feedbacks on patterns of animal and plant biodiversity^{11, 25, 26}. The result is that over large areas of Africa, vegetation structure and ecosystem biodiversity is poorly predicted by climate alone, and thus cannot be described as climate controlled²⁷, but rather strongly determined by the disturbance regime²⁸, especially under CO₂ concentrations typical of pre-industrial times and lower^{24, 29} when carbon accumulation rate and carbon allocation differentials between C₄ grasses and C₃ woody plants were accentuated³⁰. With CO₂ rising to levels³¹ last seen several millions to tens of millions of years before the pivotal evolutionary events post-Miocene³², it is difficult to see how the future of vegetation structure and biodiversity in Africa can be projected without incorporating an understanding of how disturbance regimes and climatic and atmospheric CO₂ controls will interact to determine biodiversity responses.

Unsurprisingly, there are divergent views on the future of African ecosystems and biodiversity, supported by a range of modelling and predictive approaches. Based on equilibrium assumptions of vegetation response to climate change, Africa shows among the lowest ecological sensitivity of any continent³³. In contrast, when dynamic global vegetation model (DGVM) approaches³⁴ are applied, taking into account the role of disturbance by wildfire and growth responses to changing CO₂ levels, very significant shifts in major biomes are simulated, with expansion of woody elements and reductions in grass dominance in fire-prone savannas, woodlands and grasslands³⁵, and increases in grass dominance in fire-averse semi-arid shrublands³⁶. DGVM simulations are based on a mechanistic representation of ecosystem function, and thus provide more defensible projections than correlative³⁷ analyses of biome determination^{38, 39} and species range shifts using climate drivers alone. The mechanistic DGVM projections suggest that active vegetation management⁴⁰, such as through the use of fire and mammal grazing and browsing, would be valuable elements of adaptation responses to protect both ecosystem services and biodiversity.

Long-duration observations of species range shifts in the Northern Hemisphere⁴¹ support the general projections of species range response to climate change using niche-based modelling approaches⁴², but these observations are limited mainly to regions where low temperatures limit biological activity⁸. Niche-based modelling techniques have also been applied in Africa, where water limitations to productivity are far more prevalent than low-temperature limitations⁸. These techniques project substantial biodiversity change through range reductions in indigenous and endemic species³, and large spatial shifts in geographic ranges of plants⁴³ and animals⁴⁴ that indicate significant risk to biodiversity⁶. These studies support a paradigm of individualistic species response to climate change⁴⁵, and the consequent need for passive migration and corridor spatial planning approaches to climate change adaptation⁴⁶, which are more appropriate where climate control of species range limits is strong.

In this Perspective, we briefly summarize several recent advances in understanding of how disturbance regimes, rising atmospheric CO₂ and climate could interact to affect biodiversity and ecosystem structure and function in sub-Saharan Africa (SSA). We explore the implications for understanding and projecting African ecosystem and biodiversity trajectories in a rising CO₂ world, and briefly discuss the related implications for optimal mitigation and adaptation responses (passive versus active intervention) involving natural ecosystems, and for efforts to improve predictive knowledge in this area.

African ecosystems encompass mainly tropical and subtropical climates, but with significant areas of arid and hyper-arid climatic zones between 30° and 50° of the Equator, and some representation of temperate- and Mediterranean-type climates towards its poleward latitudes. The continent is therefore home to a great diversity of ecosystems, and to a significant proportion of the world's biological diversity that remains relatively intact⁴⁷. With respect to vegetation biomes, the African land surface is covered by equatorial tropical forest ecosystems, tropical and subtropical woodlands and savannas, extensive tropical grasslands, arid shrublands and desert vegetation types, with small but biologically significant areas of Mediterranean-type ecosystems at its northern and southern fringes. There are also patches of temperate forest ecosystems in azonal sites scattered across the continent where altitude and latitude provide climatic conditions cool and wet enough to support them, or where topography affords protection from wildfire. Internal drainage of large rivers in the form of the Okavango Delta, for example, provides conditions suitable for unique azonal wetland ecosystems that contribute significantly to regional patterns of biodiversity⁴⁸. Such azonal ecosystems are not considered further here due to their limited spatial extent, but advances are being made in assessing their vulnerability to climate change through the simulation of hydrological processes⁴⁹.

Biodiversity mapping has been advancing rapidly in Africa, with improving biogeographical understanding of the spatial distribution of plants⁵⁰, mammals⁵¹, and herptile⁵² and bird diversity⁵³. This reveals that high biodiversity can be associated with regions where climate has been temporally more persistent and relatively less affected by Pleistocene climatic shifts⁵⁴. High biodiversity has also been more recently established for systems exposed to long-established disturbance regimes^{11, 18}. Thus, both the relative stability of climate and disturbance regimes has been important in maintaining high biodiversity on the African continent.

Ecological understanding of African ecosystems has benefited from an increasingly sophisticated elaboration of biome and pyrome concepts^{18, 36}, particularly in SSA, and now provides a strong basis for developing simulations of vegetation structure and function based on plant functional types⁵⁵. Performance of these simulations can be assessed using remote sensing⁵⁶, which is a recent breakthrough for the study of interactive climate and disturbance control of African ecosystems⁵⁷.

For the purposes of this Perspective, we consider three main clusters of biomes: tropical and subtropical disturbance-dependent (savanna and grassland) and disturbance-sensitive biomes (forest); arid and semi-arid water-limited shrubland and grassland ecosystems; temperate climate fire-dependent biomes (Mediterranean climate shrubland, high-altitude grasslands) and fire-sensitive biomes (temperate evergreen forest). The evidence base available, while far from complete, is sufficient to strongly suggest that climate, atmospheric CO₂ and disturbance changes are able to shift vegetation between states within and even between these clusters relatively rapidly.

Tropical and subtropical biomes. Tropical and subtropical SSA biomes comprise forest, savanna or grasslands, with grassland biomes occupying the broadest rainfall range from ±200 mm to ±3,000 mm mean annual precipitation (MAP)¹¹. The distribution of the forest biome is well predicted by climatic variables, with a very high likelihood of occurrence of MAP above 2,000 mm (ref. 58). Savannas and grasslands are poorly predicted by climate, and exist in a metastable mix where MAP lies between 250 and 1,750 mm (refs 27,59,60). The weakness of climate alone as a predictor of biome dominance in this rainfall range has been coherently recognized and explored only in the past 15 years (refs 28,60,61). R. J. Whittaker recognized a climate zone where ecosystems were poorly predicted on a temperature–precipitation plane, while the full spatial extent of these metastable biomes was first mapped globally by Bond⁹. The recognition of the interactive role of fire and of atmospheric CO₂ (ref. 30) in determining the relative dominance of non-forested biomes was closely followed by simulations that showed that palaeoecological shifts in the extent of these biomes in response to glacial–interglacial climate change could only be fully explained if the limiting effect of low atmospheric CO₂ on tree growth was accounted for in DGVMs⁶².

Recent syntheses suggest further that the establishment of fire-dependent grasslands and savannas in Africa may have occurred through entrainment of strong vegetation and even regional climatic feedbacks^{17, 63}. This implies that landscapes within a large area of SSA can experience relatively rapid shifts in vegetation structure (and thus in biodiversity), driven by changes in fire regime, climate and atmospheric CO₂ concentration^{12, URL:}